Meshing Tips/Tricks in HFSS to get correct simulation Results

Hi,
I recently modeled an UWB antenna in HFSS. After fabrication there is big deviation in measured results as compared to the simulated ones even though they show the same trend. For example,measured S11,gain and radiation pattern results differ from the simulated ones.

What are the tricks/tips to get simulated results as close as possible to the acutal results. Any tricks in terms of mesh settings, Air Box size etc etc would be helpful from the people who have hands on experience on HFSS.

If you simulate in HFSS, don't forget the vias for grounding, etc.

Can you explain it in detail ?

sorry, I have no experience in simulating antenna in HFSS, but I simuated some waveguide and microstrip in HFSS. If your circuits have some vias for signals or grounding, you should build via model in HFSS. You should make everything simulated as close as possible to the real situation.

Hi sweetchoto,

I agree with you still meshing is of great importance factor in HFSS, althoug they have a smart algorith for meshing but I think from one simulation to another the deviation seems to be not acceptable. Try to fix your mesh or to use length based meshing on surface, it helps sometimes the Meshing algorithm to start from the correct step. The sides of the air box shoud always be lambda/4 from the radiating surface.

BR

hi,
in order to be sure that there is no error with the meshing, you should check the convergence curve. If it tends toward a constant value (~0.02 for instance), there should be no errors.

Are you sure about this that if convergence curve is straight linei.e it has a constant value then there will be no errors or difference between the simulated and measured results?

Meshing is not the single possible cause of inaccuracy.

To get better correlation, you need to make the simulation model as close to the real circuit as possible. Like the actual material parameters of dielectrics and metals, finite size of the board, and discontinuity around connectors. If your model is over-simplified, you won’t see a good correlation between simulation and measurement even if you have a very dense mesh.